physical variables influencing near-shore habitat use of juvenile chinook salmon in the sacramento...
DESCRIPTION
This presentation covers key findings of a multi-year project working to evaluate the performance of levee designs on the mainstem Sacramento River. It reviews environmental factors that influence juvenile Chinook salmon habitat use at different levee sites. Project results suggest that fish more frequently occupy habitats on levees that incorporate mitigation features than levees that lack such enhancements and consist solely of large angular rock (i.e., "riprap"). Habitat characteristics that appear favorable to juvenile salmon included slightly increased water temperature, slow near-shore current velocity, and shallow depths, as well as the presence of vegetation and woody material.TRANSCRIPT
Physical Variables Influencing Near-shore Habitat Use of Juvenile Chinook Salmon in the Sacramento River
Michael Hellmair
Physical Variables Influencing Near-shore Habitat Use of Juvenile Chinook Salmon in the Sacramento River
U.S. Army Corps of Engineers
• Evaluate habitat use of focal fish species at various post-2006 levee repair sites
• Determine if on-site mitigation features are increasing habitat value to approximate “natural banks”
• Identify which microhabitat features of maximize fish use by focal species
Background & Introduction
Methods: Locations
Sampling occurred at 16 sites, encompassing 3 site categories: – Naturalized sites (n=4) – Unmitigated repair sites ( n=3) – Mitigated repair sites (n=9)
Unmitigated Repair Mitigated Repair
Naturalized
Methods: Sampling
Sampling by boat electrofishing
Methods: Sampling
Measure associated habitat characteristics at each incursion point
Methods: Sampling
Establish point-specific capture record & habitat parameters
Habitat variables: • Depth / Slope • Velocity / Gradient • Substrate • Temperature difference • Shade • Emergent woody
material • Emergent vegetation
Methods: Variables
Others: • Event • Rivermile • Site/design category
Methods: Habitat Occupancy • Single-variable logistic regression model to determine if
model fit is significantly improved by any one predictor:
• Remove non-significant (p> 0.25) predictors from scope for multi-variable model fits
• Use backward model selection to determine the most likely multivariate model
Pi =eg(x )
(1+ eg(x ) )
Methods: Habitat Occupancy • Test fit of the selected model using Hosmer-Lemeshow
goodness-of-fit statistic (a high p-value indicates a good fit)
• Evaluate classification accuracy with unbiased jackknife estimator
• Determine Cohen’s kappa statistic as a chance-corrected measure of prediction
Results: Habitat Value of Mitigated Repair Sites
• Fish densities at mitigated repair sites (all designs) were not significantly different from naturalized sites
• Fish densities at most mitigated repair sites were significantly higher than at non-mitigated sites
Results: Fry Habitat Occupancy
Variables excluded due to non-significance in single model evaluation: • Shade • Substrate (at 15 feet) Multivariate model fitting: Final model • Vegetation density • Depths close to shore (5 & 10 feet) • Velocity close to shore (5 & 10 feet) • Current gradient • Substrate close to shore (5 & 10 feet) • Rivermile
Results: Fry Habitat Occupancy
Occupancy probability key factors
Higher probability at points with: • Submerged vegetation (sparse, OR = 2.07) Lower probability at points with: • Deep water close to shore (OR = 0.63) • Faster current close to shore (OR = 0.46)
Hosmer-Lemeshow GOF: p = 0.34 Jackknife : 88% classified correctly Cohen’s kappa: 0.29 (Z = 6.54, p < 0.01)
Results: Juvenile Habitat Occupancy
Variables excluded due to non-significance in single model evaluation: • Shade • Depth (at 15 feet) Multivariate model fitting: Final model • Bank slope • Density of woody material • Depths close to shore • Current gradient • Temperature difference • Substrate • Rivermile
Results: Juvenile Habitat Occupancy
Occupancy probability key factors
Higher probability at points with: • woody material (sparse OR = 1.78,
medium OR = 2.71) • warmer ambient temperatures (OR = 1.64) Lower probability at points with: • Deep water close to shore (> 5ft, OR =
0.06) • Cooler ambient temperatures (OR = 0.45) Hosmer-Lemeshow GOF: p = 0.46 Jackknife : 81% classified correctly Cohen’s kappa: 0.27 (Z = 5.96, p < 0.01)
Results: Smallmouth Bass
Variables excluded due to non-significance in single model evaluation: • Vegetation density • Depth at 10 and 15 feet • Velocity gradient • Substrate
Multivariate model fitting: • Bank slope • Density of woody material • Nearshore current velocity • Rivermile
Higher chance of occupancy at: • Steep slopes (OR 2.78) • Density of woody material (OR, Low: 1.93, Medium: 3.06 , High: 11.11) • Velocity close to shore (Medium, OR 3.31) • Abundance decreases with distance upstream (OR 0.98)
Resident Rearing vs. Migration
• Collect drift samples at select sites • Gastric lavage of juvenile Chinook • Dissection of mortalities
Key points: • Majority of individuals had identifiable gastric contents (>95%) • Often large number of diet items (~200) suggestive of active feeding • Seasonally high abundance of larval native fishes in drift and diet
– Larval fishes can constitute > 60% of drift items
• Typically, copepods and cladocerans constitute > 90% of prey items
Questions?